ARTICLE IN PRESS

Journal of Crystal Growth 310 (2008) 4430–4437

Contents lists available at ScienceDirect

Journal of Crystal Growth

0022-02

doi:10.1

� Corr

E-m

journal homepage: www.elsevier.com/locate/jcrysgro

In-situ surface preparation of nominally on-axis 4H-SiC substrates

J. Hassan �, J.P. Bergman, A. Henry, E. Janzen

Department of Physics, Chemistry and Biology, IFM, Linkoping University, S-58183 Linkoping, Sweden

a r t i c l e i n f o

Article history:

Received 2 May 2008

Received in revised form

26 June 2008

Accepted 30 June 2008

Communicated by M. SkowronskiNormarski diffractional interference contrast and atomic force microscopy with tapping mode before

Available online 11 July 2008

PACS:

81.65.CF

81.15.GH

81.65.�b

82.40.�g

68.37.Ps

61.72.Ff

Keywords:

A1. Atomic force microscopy

A1. Etching

A3. Chemical vapor deposition processes

48/$ - see front matter & 2008 Elsevier B.V. A

016/j.jcrysgro.2008.06.083

esponding author. Tel.: +46 13 28 23 80, Mobi

ail address: jawul@ifm.liu.se (J. Hassan).

a b s t r a c t

A study of the in-situ surface preparation has been performed for both Si- and C-face 4H-SiC nominally

on-axis samples. The surface was studied after etching under C-rich, Si-rich and under pure hydrogen

ambient conditions at the same temperature, pressure and time interval using a hot-wall chemical

vapor deposition reactor. The surfaces of all the samples were analyzed using optical microscopy with

and after in-situ etching. Polishing related damages were found to be removed under all etching

conditions, also the surface step structure was uncovered and a few defect-selective etch pits were

observed. For the Si-face sample, the best surface morphology was obtained after Si-rich etching

conditions with more uniform and small macro-step height which resulted in the lowest surface

roughness. For the C-face sample the surface morphology was comparable under all etching conditions

and not much difference was found except for the etching rate. Si droplets were not observed under any

etching conditions.

& 2008 Elsevier B.V. All rights reserved.

1. Introduction

SiC is a material suitable for high-power electronic devices thatcan operate in harsh environment. The material quality, does notyet allow bulk material to be used as voltage supplying layer.Therefore, SiC epitaxy is a key technology for producing devicequality layers. Commercially available SiC wafers used forepitaxial growth still have a high density of polishing scratcheson the substrate surface [1,2]. High density of structural defects,like micropipes, dislocation clusters and voids, in the substrateproduce surface imperfections during polishing and enhance thesurface roughness [3]. Generally, both bulk and epitaxial growthare highly dependent on the surface quality of the startingmaterial [4]. The substrate surface damages can affect the epilayerquality by introducing several different kinds of extended defects[5–9].

Random nucleation of 3C polytype inclusions at the substra-te–epilayer interface is one of the major problems in the case ofepilayer growth on on-axis Si-face 4H-SiC substrate and it hasbeen reported that the substrate surface damages are preferentialnucleation sites for these inclusions [10,11]. A high surface

ll rights reserved.

le: +46 73 75 234 09

roughness of the epilayer grown on on-axis substrate is anotherproblem which could also be related to the morphology of thestarting surface. Furthermore, the epitaxial surface roughnessvalues are important in many processing steps in devicefabrication, as well as for the performance of many devices.

Therefore, surface preparation prior to the epilayer growth is acrucial step to obtain high-quality smooth homoepitaxial layers.SiC surface preparation is also crucial for the formation ofgraphene [12,13], which is a thin layer of carbon formed on thesurface after Si out diffuse. The in-situ surface preparation inhydrogen ambient, in the temperature range of 1400–1800 1C, isalready known to improve the surface morphology by removingpolishing scratches on off-cut substrates [14–18]. In order to avoidthe depletion of carbon from the substrate surface through theformation of hydrocarbon, a small amount of propane is usedalong with the carrier gas during temperature ramp-up whichultimately helps avoiding the formation of Si droplets on thesubstrate surface and is typically used as a routine in-situ surfacepreparation method [19]. An effect of adding propane to thehydrogen flow is also a slightly decreased etch rate and asmoother surface. Normal temperature used for epitaxial growthis around 1600 1C and the best etching results are reported at atemperature of 1400 1C [20]. With induction heating system ittakes 5–10 min to reach from 1400 1C to a stable growthtemperature of 1600 1C. During temperature ramp-up, the surface

ARTICLE IN PRESS

J. Hassan et al. / Journal of Crystal Growth 310 (2008) 4430–4437 4431

etching is inevitable and can effectively change the surfacemorphology. Therefore, it is important to know how the surfacelooks like prior to the start of the epilayer growth.

In-situ surface preparation under C-rich conditions (H2+C3H8)has been reported for both off- and on-axis substrates; etching underpure hydrogen conditions was shown to produce high density of Sidroplets on the surface of off-cut substrates [21]. The in-situ surfacepreparation of 81 and 41 off-cut substrate, under Si-rich conditions,has also been shown to improve the surface morphology [22]. Thein-situ surface preparation under different ambient conditions mayhave different effect on the on-axis substrates. We report here theexperimental results of the in-situ surface preparation of bothSi- and C-face on-axis samples under C-rich conditions (H2+C3H8),Si-rich conditions (H2+SiH4) and in pure hydrogen ambient.

2. Experimental procedure

For the etching study two commercially obtained nominally on-axis 200 4H-SiC wafers, one with Si-face [0 0 01] polished and onewith C-face [0 0 0 1] polished, were cut into small 1 cm2 pieces. Allsamples were cleaned using standard wafer cleaning process. Ahorizontal hot-wall CVD reactor [23] was used for the in-situetching of the samples. Hydrogen purified through heated palla-dium membrane was used as carrier gas. Etching was performed onboth Si- and C-face polished sample under C-rich, Si-rich and inpure hydrogen ambient conditions. During C- and Si-rich conditionsthe flow rates of C3H8 and SiH4, added to the 57 l/min flow rate ofhydrogen, were 25% and 20% of the flow rates normally used for SiCgrowth, respectively. The counting of the etching time was startedfrom a temperature of 1530 1C. All etching experiments wereperformed for 20 min each at a temperature of 1580 1C and apressure of 200 mbar. Optical microscopy with Nomarski inter-ference contrast and atomic force microscopy (AFM) in tappingmode were used to observe the surface morphology, step structureand surface roughness on all samples before and after etching.

3. Results and discussion

Optical images taken from as-received samples, from Si- andC-faces, are shown in Fig. 1a and b, respectively. Along with thepolishing scratches, small pin-hole like depressions marked withcircles can also be seen in both images. AFM performed on bothfaces revealed two different kinds of features on the surface,independent of the surface orientation:

(i)

shallow and deep polishing scratches, (ii) shallow and deep depressions.

Fig. 1. Optical image taken from as-received (a) Si face and (b) C f

The depth of the shallow polishing scratches as measuredthrough section analysis of the AFM image (Fig. 2a) is �3 nm,while for deep polishing scratches it is �10 nm. Figs. 2b and c arehigh-magnification images taken from the depressions markedwith arrows in Fig. 2a. These features were identified as small pin-hole like depressions with optical microscopy. The sectionanalysis of these micrographs shows that the diameters of thedepressions are �267 and �343 nm while the depth is �3.1 and�15.9 nm, respectively. These depressions could also be related tothe polishing process.

3.1. Etching under C-rich conditions

Optical images taken from the Si-face polished sample afteretching under C-rich conditions (Fig. 3a) show heavy stepbunching, which results in the formation of macro-steps andvery wide terraces on the surface. The macro-steps are neitherstraight, nor continuous. A few randomly distributed round etchpits also appeared on the surface after etching. The density of suchpits is found to be �300 cm�2 which is too low to be attributed tothe normal screw dislocations density (�3�103 cm�3) in com-mercially available 4H-SiC substrates [24]. On the other hand,optical micrographs (Fig. 3b) taken from C-face sample etchedunder the same conditions show a mirror-like surface without anydefects related etch pits. This could be related to the lower surfaceenergy on C-face compared to on Si-face [25]. The high surfaceenergy on Si-face is lowered through the mechanism of stepbunching. The polishing scratches and deep depressions in thescratched lines present on the surface of as-received samplesvanished after etching.

In order to examine the surface roughness and surface stepstructure AFM in tapping mode was performed. A wide area scanof 100�100mm2 taken from the Si-face sample (Fig. 4a) shows asurface roughness (RMS) of 17 nm. The macro-step height isbetween 60 and 90 nm and the terrace width is between 20 and40mm. The macro-steps are clearly not continuous. Along withthese macro-steps, the in-situ etching uncovered three differentkinds of features on the surface:

(i)

ace. In

shallow etch pits,

(ii) periodic train of micro-steps,

(iii)

deep round etch pits.

3.1.1. Shallow etch pits

On wide terraces a high density of shallow etch pits withhexagonal spiral-like structure were observed in AFM surfacetopograph and are marked with solid arrows in Fig. 4a. Analysis ofthe whole sample surface with AFM showed that such shallow

both images the pin-holes are highlighted with circles.

ARTICLE IN PRESS

Fig. 2. (a) 5�5mm2 AFM scan taken from the Si-face sample. Scratches from polishing and two depressions, marked with arrows, are observed (b) high magnification

image taken from small diameter depression and (c) high magnification image taken from large diameter depression.

Fig. 3. Optical image taken from (a) Si-face and (b) C-face sample etched in C-rich conditions.

J. Hassan et al. / Journal of Crystal Growth 310 (2008) 4430–44374432

etch pits are present over the entire sample surface and thattheir density is comparable to that of elementary screw disloca-tion’s density in 4H-SiC. High-resolution scans around theseshallow etch pits reveal a spiral-like feature (Fig. 4b) with micro-step height of 2–4 bilayers of 4H-SiC. Etching, in many ways,is a reverse process to the crystal growth. Similar to the crystalgrowth in layer-by-layer fashion at certain supersaturation,etching could be performed at certain undersaturation and thelayer height observed on the etched surface may be the height of asingle-crystal cell. Screw dislocations intersecting the surfaceprovide growth steps on the surface and are responsible forgrowth below a critical supersaturation [26]. During growth thestep advances and rotates in the form of a spiral and appears as ahillock on the surface. Similarly, in the case of etching, screwdislocations intersecting the surface provide steps which rotate inthe form of spirals in opposite direction and will appear in theform of etch pits on the surface. Therefore, shallow etch pitsappearing on the surface with hexagonal spiral-like feature areattributed to screw dislocations intersecting the surface normally.The step height corresponds to the magnitude and the clockwiseor anti-clockwise rotation of the spiral gives the sense of the

direction of Burger vector. Therefore, hexagonal spiral-likestructure in Fig. 4b corresponds to a screw dislocation with aBurger vector of �1c as the step height is 1 nm which correspondsto unit cell height in 4H-SiC and the spiral is moving clockwisetowards the reader [27].

The depth of such shallow pits varies a few nm over the entiresample surface. In Fig. 4b, the depth of the etch pit measured fromthe surface is about 12 nm. The steps at the interfaces of all sixfacets of the hexagon are intercepting at the facet boundary. Insome cases, the step height is 2 bilayers on each facet whichindicate that the unit cell steps dissociate into two steps with halfunit cell height. Shallow pits are associated with low etch ratewhereas deeper pits with higher local etch rate. The local etch ratemay depend on three factors: (i) local thermodynamic conditionsover the sample surface like temperature, pressure and concen-tration of C and Si containing species in the gas phase (ii) localdoping of the substrate and (iii) local orientation of the surfaceplanes or angle of the dislocation line relative to the surface. Inour case local orientation of the surface planes is probably themain reason for different local etching rates, resulting in differentdepth of the etch pits.

ARTICLE IN PRESS

Fig. 4. (a) AFM image 100�100mm2 of Si-face etched in C-rich conditions. Macro-steps are visible. (b) 50�50mm2 scan taken around shallow hexagonal etch pit and (c)

high-resolution 20�20mm2 scan taken from wide terrace, far away from the center of shallow hexagonal etch pit, showing micro-steps of half of unit cell height for 4H-SiC.

J. Hassan et al. / Journal of Crystal Growth 310 (2008) 4430–4437 4433

3.1.2. Periodic train of micro-steps

The micro-steps originating from elementary screw disloca-tions spread over the entire sample surface. Far away from thecore of screw dislocation, the micro-steps appeared in the form ofa periodic and linear train of micro-steps with step height of 2–4bilayers as shown in Fig. 4c. There was no evidence of stepbunching on the steps originating from screw dislocations eitherinside or outside of the shallow etch pits.

3.1.3. Deep round etch pits

The deep round etch pits marked with broken arrows in Fig. 4awere identified to be the same as seen in the optical image(Fig. 3a). High magnification AFM images revealed two differentstructures of these pits. The first one showed steep walls and ahexagonal hole in the middle of the bottom of the etch pit asshown in Fig. 5a. A high-magnification image taken from the etchpit is shown in the inset of Fig. 5a. The bottom is convex shapedand covered with micro-steps of unit cell height. A hexagonal holeappeared in the middle of the bottom. The hexagonal hole is too

deep for the AFM tip to reach its bottom. The diameter of the innerhexagonal hole is between 1 and 2mm, the diameter of outer etchpit is between 6 and 8mm while the height of the steep wall is�15 nm. The multiple steps of unit cell height originating from thehexagonal hole indicate that it is probably a micropipe with largec Burger vector. The second kind of deep etch pit appeared withless steep walls and a flat bottom, as shown in Fig. 5b. A high-magnification image taken around the etch pit, shown in theupper inset in Fig. 5b, revealed that the side walls are composed ofmicro-steps with very small terrace width, but still the side wallsare inclined with respect to the bottom of the etch pit. In this casethere was no hexagonal hole observed in the middle of thebottom. The shape of the bottom is convex with the center of thespiral higher than the rest of the bottom area. The diameter ofouter edges of the etch pit is between 15 and 25mm while thedepth is �10 nm. A high magnification image taken from thebottom of the etch pit, shown in the lower inset in Fig. 5b,revealed that it is covered with micro-steps of unit cell heightindicating that these steps are originating from an elementaryscrew dislocation.

ARTICLE IN PRESS

Fig. 5. AFM image taken from Si-face sample etched in C-rich conditions (a) 50�50mm2 area scan, the inset is a 10�10mm2 scan taken from the area marked with white

square (b) 100�100mm2 area scan, the top inset is a 20�20mm2 scan taken from the area marked with white square while the bottom inset is a 8�8mm2 AFM scan taken

at the bottom of the etch pit, elementary screw dislocation related spiral with unit cell step height is seen.

J. Hassan et al. / Journal of Crystal Growth 310 (2008) 4430–44374434

A 20�20mm2 surface topographic scan taken from the C-facesubstrates etched under the same conditions is shown in Fig. 6.The surface consists of very regular and periodic micro-steps. Thestep height is equal to one bilayer i.e. 2.5 A. A wavy patternperpendicular to the steps is observed all over the sample. No stepbunching or etch pits similar to the Si-face are seen on the C-facesample.

Fig. 6. 20�20mm2 AFM image taken from the C-face sample etched in C-rich

conditions. The scale on the top is in nm.

3.2. Etching under Si-rich conditions

Optical images taken from the Si-face sample after etching inSi-rich conditions also showed macro-step bunching but themacro-steps are straight, periodic and continuous resulting inevenly distributed wide terraces, as seen in Fig. 7. The density ofdeep etch pits on the surface was found to be similar to the Si-faceetched surface in C-rich conditions.

Optical micrographs taken from the C-face sample afteretching show mirror-like surface. Neither macro-steps nor etchpits are seen. All kinds of polishing related surface damages areremoved after etching.

A wide area AFM scan (100�100mm2, Fig. 8a) of an etchedSi-face has a surface roughness (RMS) of 6 nm which is 3 timesless than the surface roughness as measured on the Si-face sampleetched in C-rich conditions at the same temperature, pressure andtime interval. The macro-step height is 15–20 nm and terracewidth is 30–40mm. The smaller surface roughness, compared tothat on the Si-face sample etched in C-rich conditions, is due tosmaller and more even distribution of macro-steps. In-situ etchingunder Si-rich conditions also resulted in similar kind of featureson the surface as discussed in the previous case for Si-face sampleetched in C-rich conditions, but in this case the height of micro-steps produced by elementary screw dislocations is always equalto unit cell height of 4H-SiC while the terrace width is about135 nm. High-resolution image taken from wide terrace, given inthe inset in Fig. 8a, shows a straight periodic train of micro-steps

ARTICLE IN PRESS

J. Hassan et al. / Journal of Crystal Growth 310 (2008) 4430–4437 4435

of unit cell height. A typical AFM scan containing both shallowand deep pits is given in Fig. 8b. The deep etch pits show similarkind of morphology as observed in the case of Si-face etched inC-rich conditions. A high magnification AFM scan taken around ashallow etch pit is shown in the inset in Fig. 8b where unit cellheight micro-steps with anti-clock wise rotation of spiralcorrespond to a threading screw dislocation with 1c Burger vector.

AFM scans taken from C-face polished substrates etched in Si-rich conditions show similar kind of features as seen in the case ofC-face etched in C-rich conditions.

3.3. Etching in pure hydrogen

The surfaces etched under Si- or C-rich conditions werefinally compared to the surfaces etched with pure hydrogen.

Fig. 7. Optical image taken from the Si-face sample etched in Si-rich conditions.

Fig. 8. AFM images taken from Si-face etched sample in Si-rich conditions (a) 100�100

the wide terrace showing micro-steps and (b) 80�80mm2 area scan showing both shal

etch pit.

Optical image taken from the Si-face sample etched in purehydrogen (Fig. 9a) shows heavy step-bunching. The step densityis higher as compared to the Si-face sample etched in either Si- orC-rich conditions but the steps appeared with non-periodiczig-zag shape structure. Defect-selective deep etch pits appearedin a large number �3000 cm�2 and are comparable to the screwdislocation density in commercially available 4H-SiC substrates.This shows that almost all of the screw dislocations selectivelyetched in pure hydrogen ambient. Even though the susceptor wascoated with SiC but due to the ageing of the susceptor, graphitecan be exposed and under pure hydrogen conditions the atmo-sphere inside the susceptor could be a little biased with Ccontaining species. C-face sample does not show any features inoptical image but only a mirror-like surface.

AFM studies on this sample showed similar kind of features asthat for the Si-face sample etched either in Si- or C-richconditions. A wide area AFM scan of 100�100mm2 (Fig. 9b)shows a surface roughness of 10.7 nm. The macro-step height isvarying between 30 and 40 nm and the terrace width is between20 and 30mm. A closer look at the wide terrace surface revealsthat it is covered with micro-steps generated by elementary screwdislocation with step height of 2–4 bilayers while the terracewidth is 130–150 nm. The surface morphology of the C-facesample, as revealed by AFM was similar to the previous cases. In-situ etching under pure hydrogen ambient was also found toremove all polishing related surface damages; no Si droplets wereobserved on the surface.

3.4. Etching rate

In order to measure the etch rate 6 and 9-mm thick epilayerswere grown on both Si- and C-face substrates, respectively.The layer thickness and per consequence the growth rate werehigher on the C-face sample than on the Si-face sample, probablydue to the higher step density. Etching was performed on bothSi- and C-face grown epilayers simultaneously for each etching

mm2 area scan showing macro-steps, the inset is a 10�10mm2 area scan taken from

low and deep etch pits, the inset is a 12�12mm2 area scan taken around a shallow

ARTICLE IN PRESS

Fig. 9. (a) Optical image taken from Si-face etched in pure hydrogen. (b) AFM image (100�100mm2) of Si-face sample etched in pure hydrogen. Macro-steps are visible.

Table 1Macro-step height, micro-step height, surface roughness and etch rate on both Si-

and C-face samples under C-rich, Si-rich and pure hydrogen ambient conditions

Face C-rich conditions Si-rich conditions Pure hydrogen

Macro-step height (nm)

Si 60–90 15–20 30–40

C 0 0 0

Micro-step height (nm)

Si 0.5–1 1 0.5–1

C 2.5 2.5 2.5

Surface roughness (RMS) nm

Si 17 6 10.7

C 0.330 0.296 0.31

Etch rate (mm/h)

Si 2.26 2.34 2.23

C 1.68 2.24 1.34

J. Hassan et al. / Journal of Crystal Growth 310 (2008) 4430–44374436

condition. The etch rate was estimated through measuring epithickness before and after etching with Fourier transform infraredreflectance spectroscopy (FTIR). If we only consider the bondingstructure we would expect similar etching rates for both faces inpure hydrogen ambient. For the etching rate on the Si-face etchedunder Si-rich conditions we might instead expect a higher etchrate if a thin liquid Si film exists on the surface. The interfacialsurface free energy between a solid and a liquid is lower than theinterfacial surface free energy between a solid and a vapor.Therefore, under Si-rich conditions we would expect that theenvironment above the surface is more like interface between asolid and a liquid. Hence, a lower effective C/Si ratio on the surfacelowers the surface free energy [28] and changes the etchingmechanism. C-rich conditions, on the other hand, do not changethe surface free energy as C does not melt. Nevertheless,experimentally we found that the difference of the etching ratewas very small on Si-face sample under all etching conditions,indicating that the surface etching on Si-face under theseconditions is mainly controlled by the thermal decomposition ofthe surface Si–C bilayer due to high temperature. On the C-facesample, however, the surface etch rate was higher in Si-richconditions as compared to that in pure hydrogen or C-richconditions.

A comparison of the surface roughnesses (RMS), macro-stepheight, micro-step height and etching rates for both Si- and C-faceetched under different ambient at same temperature, pressureand time interval is given in Table 1. On Si-face samples etchedunder Si-rich conditions the macro-step height is smaller and a

low surface roughness is observed, also the micro-step height isalways equal to unit cell height.

4. Conclusions

In-situ etching in both Si- and C-rich conditions removes thepolishing related surface scratches and reveals the surface atomicsteps. The Si-face shows macro-step bunching while the C-faceshows continuous and periodic micro-steps without any stepbunching for all three etching conditions. The Si-face after etchingin Si-rich conditions shows more linear, periodic, continuous andevenly distributed macro-steps. Also, the surface roughness of theSi-face sample etched under Si-rich conditions is 3 times smallerthan after etching in pure hydrogen or C-rich conditions. Stepsoriginating from the elementary screw dislocations on the Si-facesample etched in Si-rich conditions are always of unit cell height.Therefore, Si-rich conditions can be effectively used to preparenominally on-axis Si-face sample before homoepitaxial growth. Sidroplets were not observed under any of the etching conditions.For Si-face the etch rate was found to be almost the same under alletching conditions but on C-face etching was found to be higherunder Si-rich conditions.

Acknowledgments

The authors are grateful for the financial support from NorstelAB, Swedish Governmental Agency for Innovation Systems,(Vinnova), and Swedish Energy Agency (STEM).

References

[1] C.H. Li, R.J. Wang, J. Seiler, I. Bhat, Mater. Sci. Forum 457–460 (2004) 801.[2] W. Qian, M. Skowronski, G. Augustine, R.C. Glass, D. Hobgood, R.H. Hopkins,

J. Elec. Chem. Soc. 142 (1995) 4290.[3] R.T. Bondokov, T. Lashkov, T.S. Sudarshan, Jpn. J. Appl. Phys. 43 (2004) 43.[4] R.C. Glass, D. Henshall, V.F. Tsvetkov, C.H. Carter, J. Mater. Res. Bull. (1997) 30.[5] A. Ellison, J. Zhang, J. Peterson, A. Henry, Q. Wahab, J.P. Bergman, Y.N. Makarov,

A. Vorob’ev, A. Vehanen, E. Janzen, Mater. Sci. Eng. B 61–62 (1999) 113.[6] J.A. Powell, D.J. Larkin, L. Zhou, P. Pirouz, in: Transactions of the Third

International High Temperature Electronics Conference, Sandia NationalLaboratories, Albuquerque, 1996, p. II-3.

[7] J.A. Powell, D.J. Larkin, Phys. Stat. Sol. B 202 (1997) 529.[8] J. Hassan, J.P. Bergman, A. Henry, E. Janzen, J. Appl. Phys., submitted for

publication.[9] L. Zhou, V. Audurier, P. Pirouz, J.A. Powell, J. Electrochem. Soc. 144 (1997) L161.

[10] S. Nakamura, T. Kimoto, H. Matsunami, J. Crystal Growth 256 (2003) 341.[11] J. Hassan, J.P. Bergman, A. Henry, E. Janzen, Mater. Sci. Forum 556–557

(2007) 53.

ARTICLE IN PRESS

J. Hassan et al. / Journal of Crystal Growth 310 (2008) 4430–4437 4437

[12] C. Berger, Z. Song, T. Li, X. Li, A.Y. Ogbazghi, R. Feng, Z. Dai, A.N. Marchenkov,E.H. Conrad, P.N. First, W.A. de Heer, J. Phys. Chem. B 108 (2004) 19912.

[13] A.J. van Bommel, J.E. Crombeen, A. van Tooren, Surf. Sci. 48 (1975) 463.[14] C. Hallin, F. Owman, P. Martensson, A. Ellison, A. Konstantinov, O. Kordina,

E. Janzen, J. Crystal Growth 181 (1997) 241.[15] V. Ramachandran, M. Brady, A. Smith, R. Feenstra, D. Greve, J. Electron. Mater.

27 (1998) 308.[16] A. Kawasuso, K. Kojima, M. Yoshikawa, H. Itoh, K. Narumi, Appl. Phys. Lett. 76

(2000) 1119.[17] P. Martensson, F. Owman, L.I. Johansson, Phys. Stat. Sol. B 202 (1997) 501.[18] S. Dogan, A. Teke, D. Huang, H. Morkoc- , C.B. Roberts, J. Parish, B. Ganguly,

M. Smith, R.E. Myers, S.E. Saddow, Appl. Phys. Lett. 82 (2003) 3107.[19] A.A. Burk, L.B. Rowland, G. Augustine, H.M. Hobgood, R.H. Hopkins,

in: III-Nitride, SiC and Diamond Materials for Electronic Devices Symposium(1996) 275.

[20] S. Soubatch, S.E. Saddow, S.P. Rao, W.Y. Lee, M. Konuma, U. Starke, Mater. Sci.Forum 483–485 (2005) 761.

[21] C. Hallin, F. Owman, P. Martensson, A. Ellison, A. Konstantinov, O. Kordina,E. Janzen, J. Crystal Growth 181 (1997) 241.

[22] K. Kojima, S. Kuroda, H. Okumura, K. Arai, Mater. Sci. Forum 556–557 (2007) 85.[23] A. Henry, J. ul Hassan, J.P. Bergman, C. Hallin, E. Janzen, Chem. Vapor

Deposition 12 (2006) 475.[24] A.R. Powell, R.T. Leonard, M.F. Brady, St.G. Muller, V.F. Tsvetkov, R. Trussell,

J.J. Sumakeris, H. McD. Hobgood, A.A. Burk, R.C. Glass, C.H. Carter, Mater. Sci.Forum 457–460 (2004) 41.

[25] T. Kimoto, A. Itoh, H. Matsunami, Appl. Phys. Lett. 66 (1995) 3645.[26] W.K. Burton, N. Cabrera, F.C. Frank, Nature 163 (1949) 398.[27] X. Ma, J. Appl. Phys. 99 (2006) 063513.[28] K. Kojima, S. Nishizawa, S. Kuroda, H. Okumura, K. Arai, J. Crystal Growth 275

(2005) e549.